“Data analytics may change the perspective on risk scenarios”

Thomas, the first half of the year was shaped by the return of volatility – that must have been a challenge for you.

It is true that after a very calm last year, volatility returned in February. One could even argue that we are just back to normal levels of volatility, especially if you look back longer than one year. As our risk models have strong antiprocyclicality elements, we were well prepared. In fact, the actual pick-up in volatility only reached the minimum level of volatility already implied in our models. Overall, the spikes were much lower than, for example, in 2016 around the Brexit vote, and as they met our models’ assumptions we could easily handle them. Still, it was interesting to see model behavior and to see them behave as expected.

What caught your attention when looking at model behavior?

The model behavior is monitored and validated on an ongoing basis by our Model Validation function. So, to pay attention to the model behavior is a normal process. Interesting in February was especially the interplay between the volatility-driven components of the models with the more static and flooring elements like minimum parameters, stressed period, minimum volatilities and so on.

In the beginning of the year, we saw discussions concerning the margin call as the daily market move of a number of equity index products exceeded the initial margin charged by clearing houses. What are your thoughts on this?

In general, initial margin is a very important layer in the waterfall but by far not the only one. Therefore, initial margin is calibrated to a dedicated confidence level, for example 99 percent for equity index products, and so by design there are cases where the initial margin rate is exceeded by the market move. This is also very transparently reported by every CCP in their quantitative public disclosure. On the other hand, I think the equity market moves earlier this year were not that severe and by far smaller than moves we saw in 2008 and even in 2016. In our models, we have not seen theses exceedances in the equity products as the antiprocyclicality elements of the model secured a conservative level of initial margin.

Apart from these, what other risks do you keep an eye on?

In the CCP business, there is huge focus on financial risks; however, I think it is especially non-financial risks which need to be taken care of as well. Here, information security risks are obviously a key area for attention. Particularly cyber risks require a joint industry effort by the market infrastructure providers, market participants and the regulators, and in my opinion, communication and alignment is the key aspect here. We all need to understand that these kinds of risks do not only affect a certain region, country or continent but are global risks in our network industry. This is why we need an aligned protocol covering operational, technical and legal aspects to manage these incidents.

Looking ahead into the future, what is important when it comes to risk management?

I expect changes in the risk management area due to artificial intelligence, robotics and data analytics. All of them may lead to a change in perspective on already familiar difficulties or risk scenarios. We are now able to process extremely large amounts of data very quickly, and algorithms can analyze structures data very efficiently. Consequently, we might detect dependencies and relations, which will change risk management for us as a CCP. While it will probably take some time until data analytics will be widely accepted, I strongly believe that our risk management can heavily benefit.